【摘要】 本文在讨论基本L,MS,变步长NLMs【",改进的SVSI。MS[2]和LMS／Fp]组合自适应滤波算法的上基础上提出一种新的可变步长L.MS自适应滤波算法,新算法引入修正系数p和遗忘因子／lf=exl^-i（i=l,2,...,M-1）,并利用p和^i来产生新的步长参与迭代。计算机仿真结果表明,与基本LMS算法或变步长NL,MS算法、改进的SVSLMS算法、LMS／F组合算法相比,新算法在保持算法简单这一特点的同时进一步加快了收敛速度,并能够收敛到更小且稳定的均方误差（MSE）。更多还原

【Abstract】 This paper discusses about some adapting filtering algorithms of standard LMS, variable step size NLMS, improved SVSLMS and Combined LMS/F. Then we proposed a novel adaptive filtering algorithm with variable step size. The novel algorithm introduces a real scaling factor denoted by P and forgetting factor λi=exp(-i) (i=l,2,...,M-l) to produce the new variable step size μ (n) and use theμ (n) to update the filter coefficients in each iteration. Computer simulations demonstrate that in the scenarios of channel equalization, the proposed algorithm accomplishes faster convergence and steady smaller MSE than the LMS, improved SVSLMS, NLMS and Combined LMS/F algorithms with only a small increase of the computational complexity.更多还原